Ph Herman, A Ferrant, M De Bruyère and N Straetmans

Transcription

1 Leukemia (1998) 12, Stockton Press All rights reserved /98 $ Stromal factors support the expansion of the whole hemopoietic spectrum from bone marrow CD34 DR cells and of some hemopoietic subsets from CD34 and CD34 DR cells Ph Herman, A Ferrant, M De Bruyère and N Straetmans Hematology Department, Université Catholique de Louvain, Brussels, Belgium Ex vivo expanded bone marrow CD34 DR cells could offer a graft devoid of malignant cells able to promptly reconstitute hemopoiesis after transplant. We investigated the specific expansion requirements of this subpopulation compared to the more mature CD34 and CD34 DR populations. The role of stromal factors was assessed by comparing the expansion obtained when the cells were cultured in (1) long-term bone marrow culture (LTBMC) medium conditioned by an irradiated human BM stroma (CM), (2) medium supplemented with 15% FBS (FBSM) and (3) non-conditioned LTBMC medium (LTM) for 21 days. The effect of the addition of G-CSF (G) and/or of MIP- 1 (M) to a combination of IL-3, SCF, IL-6 and IL-11 (3, S, 6, 11) was analyzed. Compared to CD34 DR cells, CD34 and CD34 DR cells gave rise to a similar number of viable cells and to a lower progenitor expansion. The expansion potential of CD34 and CD34 DR cells was equivalent in CM and in FBSM except for both the emergence of CD61 megakaryocytic cells and LTC-IC maintenance which were improved by culture in CM. In contrast, expansion from CD34 DR cells was enhanced by CM for all the parameters tested. Compared to FBSM, CM induced a higher level of CFU-GM and BFU-E expansion and allowed the emergence of CD61 cells. HPP-CFC were maintained or expanded in CM but decreased in FBSM. Compared to input, the number of LTC-IC remaining after 21 days of CD34 DR expansion culture was strongly decreased in FBSM and variably maintained or expanded in CM. Comparison with LTM indicated that stroma conditioning is responsible for this effect. G-CSF significantly improved CFU-GM and HPP- CFC expansion from CD34 DR cells without being detrimental to the LTC-IC pool. The growth of CD61 cells was significantly enhanced by G-CSF in CM. Addition of MIP-1 had no significant effect either on progenitor expansion or on LTC-IC, regardless of culture medium. We conclude that factors present in stroma- conditioned medium are necessary to support the expansion of the whole spectrum of hematopoietic cells from CD34 DR cells and to support the expansion of cell subsets from CD34 and CD34 DR. Keywords: ex vivo expansion of human bone marrow cells; stroma-conditioned medium; cytokines Introduction The past several years have witnessed increasing efforts to develop in vitro culture technology capable of generating large numbers of transplantable hemopoietic cells. 1 The results of the previously published studies in this area vary widely, depending on the stem cell source (bone marrow, cord blood, fetal liver or peripheral blood), the cell composition of the starting material (whole or mononuclear cell mixtures, selected CD34 + population or highly purified primitive cells), the combination of growth factors, the culture media or conditions and the method of analysis of expanded cells (cell surface phenotype or cell culture). 2 6 The fate of the Correspondence: P Herman, Hematology Dept, UCL/3052, Clos Chapelle-aux-champs 30, 1200 Brussels, Belgium; Fax: Received 13 May 1997; accepted 23 October 1997 primitive hemopoietic stem cell after expansion as assessed in vitro by the long-term culture initiating cell (LTC-IC) assay or by cytometric analysis is also inconstant, some investigators demonstrating maintenance and others expansion of the LTC- IC compartment compared to input values. 4 8 The same debate exists concerning the maintenance of the short- and long-term engraftment potential in mice transplant experiments More recently in humans, 14,15 hemopoietic reconstitution after non-myeloablative therapy with in vitro expanded cells was demonstrated but durable engraftment of expanded cells remains to be demonstrated. In malignancies frequently involving the CD34 + compartment such as myeloid leukemias, the CD34 + DR cell subpopulation, which has been shown to be enriched in LTC-IC, 16,17 could be a valuable starting population for expansion. 18,19 Expanded CD34 + DR cells could offer a graft depleted of malignant cells and able to provide prompt hemopoietic reconstitution, which is known to be particularly delayed in the context of autologous transplantation for acute myeloid leukemia. 20 The optimal culture conditions for expansion of these primitive cells are ill-defined and might be different from the conditions which allow expansion from the whole CD34 + population. Therefore, we analyzed the normal bone marrow (BM) CD34 + DR cell requirements for expansion in comparison with those of the more mature CD34 + and CD34 + DR + cell populations. Three points were specifically addressed. First, many authors have described in vitro expansion in stromafree medium (medium supplemented with fetal bovine serum (FBS), autologous plasma or serum-free medium) in the presence of cytokine combination. 2 7,14 However, it is known that stromal microenvironment plays a crucial role in the regulation of hemopoiesis. Verfaillie 21 has demonstrated that soluble factors secreted by the BM stroma improves LTC-IC maintenance from cultured CD34 + DR cells. We therefore evaluated if there is a role for stromal factors contained in medium conditioned by a normal BM stromal layer in ex vivo hemopoietic cell (including megakaryocytic and stem cells) expansion from CD34 + subsets by comparison with stromafree medium. Second, the choice of hematopoietic growth factors is a critical variable influencing the outcome of suspension cultures. Primitive hemopoietic progenitors are known to be unresponsive to single growth factors and require multiple cytokines to proliferate. 22 The combination of interleukin-3 (IL-3), interleukin-6 (IL-6), interleukin-11 (IL-11) and stem cell factor (SCF) has already been shown to stimulate progenitor expansion from a stem cell-enriched population. 23 Granulocyte colony-stimulating factor (G-CSF) is able to trigger the proliferation of multipotent hemopoietic progenitor cells. 24 However, when added to a cytokine cocktail, G-CSF was reported either to improve 25 or to decrease 26 progenitor expansion. We therefore analyzed the role of G-CSF on the expansion of a primitive cell population. Third, macrophage inflammatory protein-1 (MIP-1 ) is a negative regulator of hemopoiesis that shows suppressive activity on immature pro-

2 736 In vitro expansion of CD34 + cell subsets in conditioned medium genitors requiring more than one growth factor to proliferate. 27 It has been suggested that MIP-1 could enhance hemopoietic progenitor expansion from a primitive cell population by protecting these cells from terminal differentiation. 28 This molecule has also been shown to support the maintenance of LTC-IC from BM CD34 + DR cells. 29 These experiments combined MIP-1 with IL-11 and SCF 28 or with IL-3 and conditioned medium. 29 We assessed whether this beneficial effect of MIP-1 held true when combined with a cocktail of positive growth factors such as IL-3, IL-6, IL-11 and SCF. Materials and methods Cell separation BM was aspirated from the posterior iliac crests of healthy allogeneic BM donors under general anesthesia, after informed consent according to the regulations of the local ethics committee. Cells were harvested in 50 U/ml (final concentration) preservative-free heparin (Novo Nordisk, Bagsvaerd, Denmark). The mononuclear cell fraction (MNC) was collected after separation on lymphocyte separating medium (International Medical Belgium, Brussels, Belgium). The MNC fraction was enriched in CD34 + cells by immunoaffinity column (Ceprate LC Kit; CellPro Europe, Wezembeekoppem, Belgium) according to the manufacturer s instructions. The purity of the selected CD34 + cell fraction was always 60% (mean ±s.d.: 79.0 ± 13.0%, n = 11). The CD34 + -enriched cell fraction was thereafter mixed with an equal volume of Iscove s modified Dulbecco s medium (IMDM) containing 20% DMSO (Merck, Schuchardt, Germany) and 40% FBS (Gibco BRL, Life Technologies, Merelbeke, Belgium), sealed in ampoules, frozen using a controlled rate freezer and stored in liquid nitrogen until use. Before sorting, cryopreserved aliquots were thawed at 37 C and rapidly diluted in a 10 volume of IMDM supplemented with 20% FBS (Gibco) and 50 Kunitz U/ml of DNAse I (Boehringer Mannheim, Brussels, Belgium). After centrifugation, the cells were resuspended in IMDM containing 20% FBS (Gibco) and counted under a hemocytometer. For sorting, CD34-enriched cells were labelled with phycoerythrin (PE)-conjugated anti-cd34 (HPCA-2) and fluorescein isothiocyanate (FITC)-conjugated anti-hla-dr (Becton Dickinson, Erembodegem, Belgium) monoclonal antibodies (moabs). Natrium azide was discarded from moab suspensions by dialysis counter PBS. Twenty microliters of each moab were added to cells and incubated for 30 min at 4 C. Unfixed moabs were eliminated by two washes in PBS. CD34 + subpopulations were obtained by sorting on a FACStar Plus cell sorter (Becton Dickinson) equipped with an air cooled argon ion laser ILT model 5500A (Ion Laser Technology, Salt Lake City, UT, USA). Cells were selected for low to medium forward scatter and low side scatter, expression of CD34 antigen and either high or low/no HLA-DR expression (thereafter referred to as CD34 + DR + and CD34 + DR fractions). A representative example of cell gating for sorting is shown in Figure 1. Sorted cells were collected in IMDM supplemented with 50% FBS (Gibco) and 20 ng/ml SCF (Immunex, Seattle, WA, USA). Purity of the sorted fractions was assessed by cytometric analysis immediately after sorting. Purity was routinely 95% (CD34 + DR + : 98.3 ± 1.3%; CD34 + DR : 97.8 ± 1.4%, n = 11). Liquid cultures for cell expansion Liquid media: Three liquid media were compared: (1) FBS medium (FBSM) consisted of IMDM (300 mosm/ kgh 2 O) supplemented with 100 IU/ml penicillin, 100 g/ml streptomycin and 15% preselected FBS (Gibco); (2) long-term culture medium (LTM) consisted of IMDM (340 mosm/kgh 2 O) supplemented with 100 IU/ml penicillin, 100 g/ml streptomycin, 11% preselected FBS, 11% preselected horse serum (Gibco) and M hydrocortisone sodium succinate (Upjohn, Puurs, Belgium); (3) stroma-conditioned medium (CM) was produced as follows. A stromal layer was established from allogeneic normal BM. BM cells were sedimented on hydroxyethylstarch (Plasmasteril, Fresenius, Wilrijk, Belgium) for 1 h. Buffy coat was collected, cryopreserved in IMDM containing 10% DMSO and 20% FBS and stored in liquid nitrogen. After thawing, cells were inoculated into a T175 tissue culture flask (Falcon, Becton Dickinson Labware, Lincoln Park,NJ, USA) with 100 ml of LTM. At weekly intervals, half of the supernatant was removed and replaced with 50 ml of freshly prepared LTM. At confluence, the adherent layer was irradiated at 1500 cgy (3.09 ± 0.18 Gy/min for 5 min) using a Cs 137 -source irradiator (IBL 637, ORIS-I, Saclay, France). After irradiation, the culture medium was removed and the adherent layer was washed with PBS in order to eliminate remaining non-adherent cells and cell debris and 100 ml of fresh LTM was added. Every second day for 3 4 weeks, half the medium was removed and replaced by fresh medium. The medium recovered was frozen at 20 C without further manipulation. Before use, CM was thawed at room temperature and 2 mmol/l l-glutamine (Gibco) were added. CM produced by two different normal BM stromas were used for all the experiments and gave similar results. Recombinant human cytokines: Purified recombinant human (Rh) IL-3 (specific activity: U/mg) and Rh IL- 6 (specific activity: U/mg) were generously supplied by Sandoz (S Vancayzeele, Brussels, Belgium). Rh SCF (specific activity: U/mg) was a gift of Immunex (M Widmer, Seattle, WA, USA). Rh IL-11 (specific activity: U/mg) and Rh MIP-1 (maximum inhibitory activity: ng/ml) were kindly provided by Genetics Institute (L Lisher, Cambridge, MA, USA). Rh glycosylated G-CSF (specific activity: U/mg) was purchased from Chugaï-Rhône-Poulenc (Antony, France). Growth factors were used at the following optimal concentrations predetermined by clono- genic assay: 20 ng/ml IL-3, 50 ng/ml IL-6, 20 ng/ml SCF, 20 ng/ml IL-11, 50 ng/ml MIP-1 and 20 ng/ml G-CSF. Cell culture: CD34 + (post-column), CD34 + DR + and CD34 + DR (post-sorting) cells were seeded into 24-well plates (Falcon) in triplicate in 1 ml of either FBSM, CM or LTM supplemented with cytokines except in some experiments where cytokines were omitted. Four cytokine combinations were tested in parallel: IL-3+SCF+IL-6+IL-11 (3,S,6,11) ± G- CSF (G) ± MIP-1 (M). At weekly intervals for 3 weeks, cultures were demipopulated and supplemented with cytokines and fresh medium. Cells recovered weekly from triplicate experiments were pooled and washed in IMDM containing 2% FBS (Gibco). Viable cells were counted using the trypan blue dye exclusion test and evaluated for their content in different progenitor subsets (see below). Cultures were maintained in a humid atmosphere at 37 C with 5% CO 2 in air.

3 In vitro expansion of CD34 + cell subsets in conditioned medium 737 Figure 1 Representative example of CD34 + DR + and DR cell sorting gates. Sorting was performed on CD34-enriched post column cells with low SSC and medium to low FSC in order to exclude residual contaminating CD34-negative cells. Anti-CD34-PE and anti-hla-dr-fitc were used. (a) R1; CD34 + cells; X-axis: SSC, Y-axis: FSC. (b) R2: Expression of HLA-DR vs CD34 on gated CD34 + BM cells. X-axis: CD34 (orange fluorescence), Y-axis: HLA-DR (green fluorescence). (c) Sorting gates on R2: X-axis: CD34-PE, Y-axis: HLA-DR-FITC. R3: CD34 + DR cells (25%) and R4: CD34 + DR + cells (75%). The following formula was used to calculate progenitor expansion in liquid culture at days 7, 14 and 21: (No. colonies per plate) (No. viable TNC per well) (No. cells per plate) (initial No. colonies per well) For the CD34 + cell fraction, expansion calculations were corrected according to the purity of the starting population. In order to exclude that some primitive progenitors adhered to the well bottom after 21 days of culture, the presence of plastic-adherent progenitor cells was assessed according to the previously described procedure. 30 The wells were emptied and washed twice with PBS. One milliliter of LTM supplemented with 10% 5637 bladder carcinoma-conditioned medium was added. After 1 week of culture, the supernatant was collected, cells were centrifuged and plated in clonogenic assay in agar. CFU-GM were enumerated after 14 days of semi-solid culture. Analysis of the starting and expanded cell populations The different cell subfractions were evaluated before expansion culture initiation and at weekly intervals for 3 weeks during expansion culture. Their content in CFU-GM, BFU-E, HPP-CFC was assessed by clonogenic assays, and their content in CD34 cell subsets by flow cytometry. LTC-IC were semi-quantified initially and after 21 days of expansion culture by double-stage long-term bone marrow culture (LTBMC). The megakaryocytic lineage was evaluated at weekly intervals by immunocytological staining on cytospins using anti-cd61 antibody. Clonogenic assays: CFU-GM were enumerated in triplicate in agar (Difco Laboratories, Detroit, MI, USA) using 35 mm 2 plates (Nunc, InterMed, Roskilde, Denmark) as previously described. 20 BFU-E and HPP-CFC were scored in methylcellulose (Methocult GF, H4435; Stemcell Technologies, Vancouver, Canada) according to the manufacturer s instructions. Cells were plated in duplicate in 0.9% Iscove s methylcellulose containing 20 ng/ml IL-3, 20 ng/ml IL-6, 50 ng/ml SCF, 20 ng/ml GM-CSF, 20 ng/ml G-CSF and 3 U/ml EPO. Initial CD34 +, CD34 + DR + and CD34 + DR cells were plated at a cell concentration of cells/ml for clonogenic assay in agar and of cells/ml for clonogenic assay in methylcellulose. Expanded cells were plated at the

4 738 In vitro expansion of CD34 + cell subsets in conditioned medium same cell concentration multiplied by the level of total number of nucleated cells (TNC) expansion. Cultures were incubated in a humidified atmosphere at 37 C with 5% CO 2 in air for days. Colonies containing more than 50 cells were scored after 14 days of culture and classified as CFU- GM and BFU-E. HPP-CFC were scored as dense colonies, with diameters greater than 0.5 mm, after 28 days of culture. 6 LTC-IC semi-quantification: Cultures were established and maintained according to the previously described methods. 16 Stromal feeder layers were established as for stroma CM, except that T25 flasks (Falcon, Becton Dickinson Labware) containing 8 ml of LTM were used. After irradiation at confluence, adherent cells were detached by a 10 min exposure to 0.25% trypsin (Gibco) at 37 C. After washing in IMDM supplemented with 20% FBS (Gibco), the cells were transferred into six-well plates (Falcon) at a cell concentration of cells/well in 5 ml LTM. Twenty-four hours later, the culture medium was removed, the feeder layers were washed once with PBS and LT co-cultures were initiated in duplicate. In order to avoid inter-stroma variations, the stroma from the same BM donor after the same length of culture was used to assess LTC-IC initially and after 21 days of liquid culture. For initial cell population, test cells were seeded on stromal feeder layers. After the 21 day expansion culture, cells were seeded at a cell concentration proportional to the level of TNC expansion. Every week for 5 weeks, half of the supernatant was removed and replaced by 2.5 ml of freshly prepared LTM. After 5 weeks of co-culture, supernatant and adherent cells (obtained by trypsinisation of the adherent layers) were plated in clonogenic assay in agar. The number of secondary CFU-GM was used as an estimate of the number of LTC-IC present initially or after 21 days of expansion culture. The results of day 21 LTC-IC are expressed in percentage of initial LTC-IC. Flow cytometry analysis: Starting and expanded cell fractions were analyzed for the expression of CD34 and HLA-DR antigens by a two-color fluorescence analysis using PE or FITC-conjugated anti-cd34 (HPCA-2) and FITC-conjugated anti-hla-dr moabs (Becton Dickinson). PE-conjugated anti- IgG 1 and FITC-conjugated anti-igg 2a (Becton Dickinson) were used as isocontrols. Twenty microliters of each moab were added to cells and incubated for 30 min at 4 C. Unfixed moabs were eliminated by two washes in PBS. Analysis was carried out on a flow cytofluorometer (Elite; Coulter Corporation, Hialeah, FL, USA) equipped with an argon laser operating at 488 nm, 15 mw, with filter for reemission of FITC and PE. Each sample acquisition was designed to analyse at least 10 4 events. Forward scatter, side scatter and fluorescence were evaluated using logarithmic amplifiers. The proportion of CD34 + and HLA-DR + cells was assessed and further analyzed using the Cyto 3.0 software (Coulter). fixed in acetone (Vel, Leuven, Belgium):methanol (Vel):formaldehyde 37% (Merck, Darmstadt, Germany) (19:19:2) for 90 s and then transferred directly to Tris buffer saline (TBS). Cytospins were incubated with human AB serum (ABS) for 30 min and washed in TBS for 5 min. Antibodies used further in this procedure were diluted in TBS + 10% ABS. Slides were incubated for 30 min with anti-cd61 moab (Dako, Glostrup, Denmark) diluted 1/100 and transferred in TBS for 5 min. Rabbit anti-mouse Ig (Dako) diluted 1/50 was thereafter applied for 30 min. After a washing step in TBS for 5 min, slides were incubated for 30 min with the APAAP complexes (Dako) diluted 1/50 and washed in TBS. A bridging step was performed by repeating the last two steps for 10 min each. Development of the alkaline phosphatase reaction was performed by adding a filtered mixture of Fast Red TR (Sigma, Bornem, Belgium) and Naphtol-AS-MX (Sigma) salt for 20 min in the dark. Slides were then washed for 5 min in TBS followed by water. Slides were counterstained with haematoxylin Harris (Sigma Diagnostics, St Louis, MO, USA), washed in water and finally mounted in aqueous mounting medium (Aquamount, BDH, Poole, UK). Negative controls consisted of normal human platelets (platelet-enriched plasma) incubated with ABS at the first and second step of the APAAP procedure or cells known to be negative for CD61 expression. Positive controls consisted of normal human platelets or megakaryocytic leukemia cells. Megakaryocytic lineage cells were scored according to the number of CD61-positive cells per 200 cells at 100 magnification, after examination of the entire cyto- spin area. Data analysis The normality of the distribution was assessed for each parameter. Paired two-tailed Student s t-test with = 0.05 was used to compare different culture conditions within the same cell subset. Unpaired two-tailed Student s t-test with = 0.05 was used to compare the expansion data of different cell subsets in the same culture condition. Wilcoxon signed-rank test was used when normal distribution of the data was rejected. Results are presented as mean ± standard deviation. Statistical analyses were performed using NCSS software. Results Cloning efficiency of the starting CD34 + cell subpopulations Cryopreservation procedure significantly increased the mean purity of CD34 + cells by 10.7 ± 11.3%. Initial values of CFU- GM, BFU-E, HPP-CFC and LTC-IC/ cells for the different cell subpopulations are shown in Table 1. The CD34 + DR fraction was significantly depleted of CFU-GM compared to the CD34 + and the CD34 + DR + fractions. No significant difference was observed for BFU-E and HPP-CFC content between the three CD34 + cell fractions. As expected, the CD34 + DR cell fraction was significantly enriched in LTC-IC compared to the CD34 + and the CD34 + DR + fractions. Immunocytological staining: APAAP technique: Cell fractions were analyzed for their content in CD61 + cells by the alkaline phosphatase-anti-alkaline phosphatase (APAAP) method. 31 Cytospins of to cells were dried overnight at room temperature, wrapped in foil paper and frozen at 20 C until use. Thawed cytospins were initially TNC and progenitor expansion from CD34 + DR cells is significantly improved in CM when compared to FBSM CD34 + DR cells were cultured in parallel in FBSM and in CM supplemented with four cytokine combinations (3,S,6,11 ± G

5 In vitro expansion of CD34 + cell subsets in conditioned medium Table 1 Initial numbers of progenitor/10 4 cells from the different cell fractions 739 Cell fraction CFU-GM BFU-E HPP-CFC LTC-IC CD ± ± ± ± 137 CD34 + DR ± ± ± ± 102 CD34 + DR 232 ± 134* 636 ± ± ± 334** CD34 + (post-column), CD34 + DR + and CD34 + DR (post-sorting) cells were plated in CFU-GM, BFU-E, HPP-CFC and LTC-IC assays and enumerated as described in the Materials and methods section. Data are expressed as mean number of colonies/10 4 cells plated ± s.d. of 11 separate experiments. P values were calculated using two-tailed Student s t-test for unpaired samples. *Significantly lower than CD34 + and CD34 + DR + fractions. **Significantly higher than CD34 + and CD34 + DR + cell fractions. and ± M). TNC and progenitor expansion was analyzed at weekly intervals and compared to input. As shown in Table 2, CM induced a significantly higher level of viable cell amplification than FBSM for the whole culture timescale, except at day 21 in the presence of G-CSF. In both media, the number of TNC produced continued to increase until the end of the culture period. Addition of G-CSF to 3,S,6,11 significantly increased TNC counts in both media for the 3 weeks of culture. Addition of MIP-1 did not significantly affect TNC counts, regardless of culture condition. As shown in Figure 2, CM gave rise to higher CFU-GM expansion than FBSM. In FBSM, CFU-GM expansion peaked at day 14 (maximal 32-fold expansion at day 14) and declined thereafter. In CM, CFU-GM expansion increased significantly from day 7 to day 14 and plateaued thereafter (maximal 70- fold expansion at day 21). Addition of G-CSF significantly improved CFU-GM expansion regardless of the medium. Addition of MIP-1 did not significantly improve CFU-GM expansion either in CM or in FBSM (data not shown). BFU-E progressively decreased in FBSM but were maintained or slightly expanded in CM during the culture period (Figure 3). A maximal three-fold expansion was observed in CM at day 7. G-CSF and MIP-1 had no significant effect on the erythroid compartment. HPP-CFC maximal expansion was observed at day 21 in CM (six-fold expansion with 3,S,6,11 + G). HPP-CFC were maintained or expanded in CM but decreased in FBSM. Addition of G-CSF significantly improved HPP-CFC expansion Figure 2 The effect of medium on CFU-GM expansion from CD34 + DR cells CD34 + DR cells were cultured for 21 days in FBSM or in CM supplemented with (1) 20 ng/ml IL-3, 20 ng/ml SCF, 50 ng/ml IL-6, 20 ng/ml IL-11 or (2) the same cytokine combination +20 ng/ml G-CSF. CFU-GM were counted weekly by clonogenic assay in agar. Results are expressed as mean ± s.d.-fold CFU-GM expansion compared to input (8 11 separate experiments). P values were calculated using two-tailed Student s t-test for paired samples. *Significantly higher than FBSM with the same cytokine combination. at days 14 and 21 of culture (Figure 4). MIP-1 impaired HPP- CFC expansion but this effect was not statistically significant. The percentage of initial input LTC-IC remaining after 21 days of liquid culture, as assessed by semi-quantitative assay (Table 3), showed high inter-sample variation. Nevertheless, LTC-IC were strongly decreased or undetectable after expansion culture in FBSM while they were variably maintained or increased after expansion culture in CM. The difference between the two media is statistically significant. It has to be noted that in our culture conditions, even in CM, LTC-IC expansion occurred rarely (1/7 experiments). Addition of G- CSF or MIP-1 did not significantly modify LTC-IC maintenance regardless of the medium. To exclude that the decrease in LTC-IC after 21 days of expansion culture was due to the fact that these primitive cells adhered to the well bottom and were therefore not recovered Table 2 TNC expansion from CD34 + DR cells Cytokine combination FBSM Day of culture CM d7 d14 d21 d7 d14 d21 3,S,6,11 5 ± 2 28 ± ± ± 11* 111 ± 52* 400 ± 170* 3,S,6,11 + M 4± 3 24± 8 65 ± 36 8 ± 3* 69 ± 31* 338 ± 247* 3,S,6,11 + G 9± ± ± ± 10* 351 ± 201* 660 ± 288 3,S,6,11 + G + M 7± ± ± ± 4* 228 ± 115* 565 ± CD34 + DR cells were cultured for 21 days in FBSM or in CM supplemented with 20 ng/ml IL-3, 20 ng/ml SCF, 50 ng/ml IL-6, 20 ng/ml IL-11 ± 20 ng/ml G-CSF ± 50 ng/ml MIP-1 (3,S,6,11,G,M). Results are expressed as mean ± s.d.-fold TNC expansion compared to input (8 11 independent experiments). P values were calculated using two-tailed Student s t-test for paired samples. *Significantly higher than FBSM with the same cyotkine combination.

6 In vitro expansion of CD34 + cell subsets in conditioned medium 740 Figure 3 BFU-E expansion from CD34 + DR cells CD34 + DR cells were cultured for 21 days in FBSM or in CM supplemented with 20 ng/ml IL-3, 20 ng/ml SCF, 50 ng/ml IL-6, 20 ng/ml IL ng/ml G-CSF (3,S,6,11,G). BFU-E were counted weekly by clonogenic assay in methylcellulose. Results are expressed as mean ± s.d.-fold BFU-E expansion compared to input (8 11 separate experiments). P values were calculated using two-tailed Student s t- test for paired samples. *Significantly higher than FBSM with the same cytokine combination. for analysis, the presence of plastic-adherent progenitor cells was assessed according to Gordon s procedure 30 (n = 3). No plastic-adherent primitive cell could be found in any of our culture conditions. The effect of CM is due to the conditioning on stroma and not to the components of the LTM CM is different from FBSM not only by the conditioning on a stromal layer but also by the presence of horse serum and hydrocortisone. In order to assess the role of stromal conditioning, expansion in CM and in fresh LT non-conditioned medium (LTM) was compared. For the 3 weeks of culture, TNC and CFU-GM expansion was significantly lower in LTM than in CM, regardless of the cytokine combination. BFU-E expansion was also lower in LTM, but the difference did not reach statistical significance. HPP-CFC expansion was significantly higher in CM than in LTM only at day 21 of culture. Results of expansion in the two media at day 21 are shown in Table 4. LTC-IC were undetectable after 3 weeks of liquid culture in LTM. CD34 + and CD34 + DR + cell subsets are less dependent on culture medium than CD34 + DR cells The growth of CD34 + and CD34 + DR + populations was comparable for the whole culture period and for all the parameters Figure 4 HPP-CFC expansion from CD34 + DR cells CD34 + DR cells were cultured for 21 days in FBSM or in CM supplemented with 20 ng/ml IL-3, 20 ng/ml SCF, 50 ng/ml IL-6, 20 ng/ml IL-11 ± 20 ng/ml G-CSF (3,S,6,11 ± G). HPP-CFC were counted weekly by culture in methylcellulose for 28 days. Results are expressed as mean ± s.d.-fold HPP-CFC expansion compared to input (7 10 separate experiments). P values were calculated using twotailed Student s t-test for paired samples. Comparison of HPP-CFC expansion after culture in FBSM or in CM (1) +3,S,6,11 and (2) +,S,6,11 + G. *Significantly higher than FBSM with the same cytokine combination. **Significantly higher than CM1. Table 3 Percentage of input LTC-IC remaining after 21 days of CD34 + DR expansion culture Cytokine combination FBSM CM P 3,S,6,11 1 ± 3 19 ± ,S,6,11 + G 3 ± 4 61 ± ,S,6,11 + M 2 ± 4 76 ± ,S,6,11, + G + M 3 ± 6 45 ± CD34 + DR cells were cultured for 21 days in FBSM or in CM supplemented with 20 ng/ml IL-3, 20 ng/ml SCF, 50 ng/ml IL- 6, 20 ng/ml IL-11 ± 20 ng/ml G-CSF ± 50 ng/ml MIP-1 (3,S,6,11,G,M). After 21 days of culture, cells were plated in LTC- IC assay for 5 weeks. The number of secondary CFU-GM present in the supernatant and in the adherent layer after 5 weeks of LT culture was used as an estimate of the LTC-IC. Results are expressed as mean percentage ± s.d. of input LTC-IC remaining after 21 days of liquid culture of CD34 + DR cells (4 7 separate experiments). P values were calculated using two-tailed Student s t-test for unpaired samples. tested. CM and FBSM gave rise to a similar TNC amplification with a maximum at day 21 (FBSM: CD34 + DR + : 398 ± 148 vs CD34 + : x375 ± 163; CM: CD34 + DR + : 364 ± 80 vs CD34 + : 431 ± 163, n = 7). In terms of CFU-GM, BFU-E and HPP-CFC expansion, there was no statistical difference between CD34 + and CD34 + DR + subpopulations and no difference between the two media regardless of cytokine cocktail for the whole culture period. Results of CFU-GM expansion are shown in Table

7 In vitro expansion of CD34 + cell subsets in conditioned medium Table 4 LTM Comparison of CD34 + DR cell expansion in CM and in 741 Culture condition Fold expansion at day 21 of culture TNC CFU-GM BFU-E HPP-CFC CM: 3,S,6, ± 144* 25 ± 8* 0.9 ± ± 3.1* CM: 3,S,6,11 + G 791 ± 323* 39 ± 8* 2.6 ± ± 0.6* LTM: 3,S,6,11 72 ± 71 3 ± ± ± 0.2 LTM: 3,S,6,11 + G 239 ± ± ± ± CD34 + DR cells were cultured for 21 days in LTM or in CM supplemented with 20 ng/ml IL-3, 20 ng/ml SCF, 50 ng/ml IL- 6, 20 ng/ml IL-11 ± 20 ng/ml G-CSF (3,S,6,11,G). Results are expressed as mean ± s.d.-fold expansion compared to input values after 21 days of liquid culture (4 6 separate experiments). P values were calculated using two-tailed Student s t-test for paired samples. *Significantly higher than LTM with the same cytokine combination. 5. Semi-quantitative assessment of LTC-IC content after 3 weeks of liquid culture showed an LTC-IC decrease compared to input values in the presence of 3,S,6,11 + G in FBSM and in CM (FBSM: CD34 + DR + :0± 0% vs CD34 + :3± 3% and CM: CD34 + DR + :14± 20% vs CD34 + :11± 5%, n = 2 5). However, the maintenance of LTC-IC from CD34 + cells was significantly better when the cells were cultured in CM rather than in FBSM. Regardless of culture condition, CD34 + DR + cells showed higher TNC amplification than CD34 + DR cells at day 7 (CD34 + DR : 19 ± 11 vs CD34 + DR + : 66 ± 20, n = 8). This difference disappears at day 14 (CD34 + DR : 322 ± 186 vs CD34 + DR + : 344 ± 94, n = 7) and at day 21 of culture (CD34 + DR : 621 ± 230 vs CD34 + DR + : 516 ± 440, n = 7). In CM, CFU-GM expansion from CD34 + DR cells was significantly higher than that of CD34 + DR + cells for the whole culture period. In FBSM, this difference is significant only at day 14. CFU-GM expansion from CD34 + DR + cells was not dependent on culture medium contrary to CD34 + DR cells (Figure 5). The same observations held true regarding the comparison between CD34 + and CD34 + DR cell fractions. CD34 + and CD34 + DR + but not CD34 + DR cells survived in CM without added cytokines In order to evaluate if CM by itself contains all the factors required for cell growth and survival, CD34 +, CD34 + DR + and Figure 5 Comparison of CFU-GM expansion from CD34 + DR + and DR cells cells were cultured for 21 days in FBSM or in CM supplemented with 20 ng/ml IL-3, 20 ng/ml SCF, 50 ng/ml IL-6, 20 ng/ml IL-11 and 20 ng/ml G-CSF. CFU-GM were counted weekly by clonogenic assay in agar. Results are expressed as mean ± s.d.-fold CFU-GM expansion compared to input (8 11 separate experiments). P values were calculated using two-tailed Student s t-test for unpaired samples. *Significantly higher than CD34 + DR + in the same culture condition. CD34 + DR cells were cultured in CM or in FBSM without exogenous cytokines. In CM, CD34 + and CD34 + DR + cells survived at least 3 weeks. During this period, a slight increase in TNC (mainly macrophagic cells) (CD34 + DR + : 12 ± 8 at day 21, n = 4) was observed but the clonogenic potential of these cells was completely lost (n = 3). Cells died beyond day 7 of culture in FBSM. In contrast, CD34 + DR cells died rapidly (after 2 or 3 days) in either CM or FBSM when no exogenous cytokines were added (n = 3). CD61 + cell expansion is improved in CM and in the presence of G-CSF The megakaryocytic lineage was analyzed by immunocytological staining performed on cytospins of cultured cells using anti-cd61 antibody. In order to exclude that CD61 + cells seen during liquid culture arose from the survival of megakaryocytes (MK) present at the time of culture initiation, APAAP immunostainings were performed on initial cell fractions. Table 5 Comparison of CFU-GM expansion from CD34 + DR + and CD34 + cells Day of culture CD34 + DR + CD34 + d7 d14 d21 d7 d14 d21 FBSM: 3,S,6,11 5 ± 3 8 ± 4 8 ± 5 5 ± 2 8± 5 6± 4 FBSM: 3,S,6,11 + G 6 ± 5 16± ± 16 5 ± 4 11± 7 8± 9 CM: 3,S,6,11 7 ± 3 11± 6 14± 8 5 ± 2 9± 5 9± 6 CM: 3,S,6,11 + G 6 ± 3 16± ± 25 6 ± 3 12± 8 12 ± CD34 + or CD34 + DR + cells were cultured for 21 days in FBSM or in CM supplemented with 20 ng/ml IL-3, 20 ng/ml SCF, 50 ng/ml IL-6, 20 ng/ml IL-11 ± 20 ng/ml G-CSF (3,S,6,11,G). Results are expressed as mean ± s.d.-fold CFU-GM expansion compared to input values after 21 days of liquid culture (6 7 separate experiments). P values were calculated using two-tailed Student s t-test for unpaired samples.

8 742 In vitro expansion of CD34 + cell subsets in conditioned medium Neither morphologically identifiable MK nor immature CD61 + cells were detectable in the starting CD34 +, CD34 + DR + and CD34 + DR cell fractions (n = 5). On morphological examination of CD61 + cells appearing during expansion, mature MK were rarely observed. In our culture conditions, CD61 + cells appeared predominantly as round-shaped cells with high to medium nucleocytoplasmic ratio, indicating cytoplasmic maturation without nucleus polyploidization (Figure 6). Compared to FBSM, CM significantly improved the emergence and expansion of CD61 + cells (Table 6). In CM, the number of CD61 + cells significantly increased from day 7 to day 14 of culture and not significantly thereafter. These observations held true for the three cell subsets. For CD34 + DR cells, the highest number of CD61 + cells was observed at day 21 of culture in CM supplemented with 3,S,6,11 + G. Addition of G-CSF significantly improved CD61 + cell expansion compared without G-CSF (day 7: 2, day 14: 4 and day 21: 4) in CM from the CD34 + DR cell subpopulation only. Addition of MIP-1 did not improve CD61 + cell numbers, either in FBSM or in CM (data not shown). a b Figure 6 Representative examples of CD61 + cells. CD34 + were cultured in CM supplemented with 20 ng/ml IL-3, 20 ng/ml SCF, 50 ng/ml IL-6, 20 ng/ml IL-11 and 20 ng/ml G-CSF. Cytospin of cultured cells were stained with the APAAP procedure using anti-cd61 antibody as described in the Materials and methods section. (a) Mononucleated CD61 + cell. (b) Binucleated CD61 + cell. CD34 antigen expression rapidly decreases during ex vivo expansion regardless of medium, cytokine combination or CD34 + subpopulation Cytometric analyses showed a rapid decrease in CD34 antigen expression during culture regardless of cell fractions or culture conditions. Average CD34 + percentage was 7.0 ± 5.0% at day 7, 1.5 ± 1.6% at day 14, and 1.0 ± 0.3% at day 21 of liquid culture (n = 5). The percentage of CD34 + decreased in culture while TNC number increased. Consequently, the absolute number of CD34 + cells was at least maintained. HLA-DR expression increased rapidly from day 0 to 7 of culture of CD34 + DR cells and plateaued thereafter (CD34 DR + : day 7: 23 ± 8%, day 14: 33 ± 16% and day 21: 32 ± 13%, n = 5). Discussion In this study, we demonstrated that CM supplemented with exogenous cytokines is necessary to support the expansion of myeloid and erythroid progenitors, and also to support the emergence of megakaryocytic cells and the partial maintenance of LTC-IC from CD34 + DR cells. In contrast, culture of these primitive cells in medium containing FBS without stromal conditioning induces poor progenitor expansion and almost no HPP-CFC or LTC-IC maintenance. By comparison with the expansion of CD34 + DR cells in non-conditioned LT medium, we show that stromal conditioning is responsible for the effect of CM. However, although indispensable for CD34 + DR expansion, CM is not sufficient by itself and requires the addition of exogenous cytokines. Stroma-conditioned medium had already been shown to support the maintenance of LTC-IC. 21 The diffusible factors responsible for the observed effect could involve cytokines produced by the stroma and/or extracellular matrix components. BM stromal and endothelial cells are known to produce in picogram concentrations a variety of cytokines (such as M-CSF, G-CSF, GM- CSF, SCF, TGF-, IL-1, IL-6, LIF and IL-11) involved in the regulation of the survival, proliferation and differentiation of hemopoietic progenitor and stem cells. 32,33 Recently, Flt3- ligand 34 and thrombopoietin 35 have also been shown to be produced by human BM stromal cells. Extracellular matrix components are produced by both hemopoietic and stromal cells. 36 Recently Gupta et al 37 demonstrated that heparan sulphate was involved in the LTC-IC maintenance induced by stroma-conditioned medium. The impact of culture medium on cell expansion and survival varied according to the cell subpopulation analyzed. CD34 + and CD34 + DR + cell fractions gave comparable results in expansion. TNC, CFU-GM, BFU-E and HPP-CFC from these two subpopulations are equally expanded or maintained whether they are cultured in FBSM or in CM. This contrasts with the previously discussed dependence of CD34 + DR cells on CM for expansion. However, CM was not completely devoid of effect on these CD34 + and CD34 + DR + cells. In the absence of exogenous cytokines, CM improved the survival of CD34 + and CD34 + DR + cells compared to FBSM. CM was necessary for CD61 + cell emergence. The LTC-IC recovery after a 21-day culture of CD34 + was significantly better in CM than in FBSM. These observations underscore the fact that expansion conditions are not universal: they vary according to the starting population (CD34 + or more primitive subsets) and according to the cell type that is to be obtained (LTC-IC, progenitor or post-progenitor cells). In apparent contrast with

9 In vitro expansion of CD34 + cell subsets in conditioned medium Table 6 Emergence of CD61 + cells in liquid culture 743 Day of culture FBSM CM d 7 d 14 d 21 d 7 d 14 d 21 CD34 + 3,S,6,11 2 ± 3 10 ± 29 0 ± 0 73± 104* 430 ± 411* 1022 ± 868* 3,S,6,11 + G 11± ± ± ± 208* 661 ± 776* 795 ± 389* CD34 + DR + 3,S,6,11 5 ± 6 1± 2 2± 4 82 ± 54* 243 ± 120* 558 ± 98* 3,S,6,11 + G 13± ± 22 6 ± ± 103* 628 ± 602* 842 ± 392* CD34 + DR 3,S,6,11 0 ± 0 0± 0 0± 0 30 ± 25* 151 ± 127* 233 ± 207* 3,S,6,11 + G 2± 3 10 ± ± ± 35* 480 ± 548* 1238 ± 759* CD34 +, CD34 + DR + and CD34 + DR cells were cultured for 21 days in FBSM or in CM supplemented with 20 ng/ml IL-3, 20 ng/ml SCF, 50 ng/ml IL-6, 20 ng/ml IL-11 ± 20 ng/ml G-CSF (3,S,6,11,G). The percentage of CD61 + cells was evaluated by counting 200 cells on weekly cytospins after immunostaining using the APAAP technique and anti-cd61 moab. The absolute count of CD61 + cells was calculated by multiplying the percentage of CD61 + cells by TNC. No CD61 + cells were detectable at culture initiation. Results are expressed as mean number of CD61 + cells 10 2 ± s.d. (5 9 separate experiments). P values were calculated using two-tailed Student s t-test for paired samples. *Significantly higher than FBSM for the same CD34 + subpopulation. our data, Breems et al 38 recently showed that CM improved in vitro progenitor expansion from total CD34 + cells. Although direct comparison is difficult because of different source of cells (PBSC vs BM) and different source of CM (stromal cell line vs BM stroma), it seems that our control medium (containing 15% FBS) gives better progenitor expansion than medium containing 2% horse serum as used by Breems et al. CM in both settings gives comparable results in terms of progenitor expansion. Taken together these data are in favor of the use of CM to improve hemopoietic cells ex vivo expansion. Although clearly better than FBSM in this respect, CM rarely produced LTC-IC expansion after 21 days of culture of CD34 + DR cells. More often, maintenance or partial loss of LTC-IC was observed, in contrast to the results of other studies. 8,39 In the present study, LTC-IC were evaluated by a semiquantitative assay, that assess both the total number of LTC- IC and the number of CFC produced by LTC-IC. Therefore we cannot exclude that expansion culture induced a decrease in the number of CFU-GM produced by LTC-IC. Although generalization is hazardous, we could suggest that this hypothesis is not supported by the data of Petzer et al 8 showing that average CFC output/ltc-ic remains unchanged after a 10 day expansion culture period. Another hypothesis is that prolonged culture may have led to depletion of LTC-IC through terminal differentiation and that the LTC-IC peak occurs before day 21. In support of this hypothesis, Verfaillie and Miller 39 showed that LTC-IC expansion from BM CD34 + DR cells occurs at week 2 of culture but that LTC-IC were only maintained and not expanded after 5 8 weeks of culture. Brandt et al 6 also showed that HPP-CFC expansion from CD34 + DR CD15 cells peaked at day 15 of culture. In contrast, our experimental data showed that, in CM, HPP-CFC expansion from CD34 + DR cells increases until day 21. Other recently published studies 8 demonstrated an 30-fold expansion of LTC-IC after 10 days of culture, and an 50-fold expansion by the end of another 1 3 weeks. Differences in starting cell population, in cell concentrations at culture initiation, in cytokine combinations (in particular the use of Flt3 ligand), in culture media and in models used to analyze primitive cells hamper definite conclusion. That the LTC-IC peak occurred before day 21 in our culture conditions remains possible. The suboptimal recovery of LTC-IC might also be due to the presence of accessory cells in the culture. Koller et al 40 suggested that sorted CD34 + lin cells from expanded material give better LTC-IC evaluation and recovery than unsorted cells. These data suggest that accessory cells could act as inhibitors for LTC-IC evaluation in ex vivo expansion protocols. In accordance with Bridell et al, 41 initial cell fractions showed neither morphologically identifiable MK nor immature CD61 + cells as assessed by APAAP immunostaining. These observations suggest that CD61 + cells observed from day 7 to day 21 of liquid culture arise from a CD61-negative starting cell population. CD61 + cell emergence was observed in CM supplemented with cytokines regardless of the subpopulation analyzed. In FBSM, cytokines are not sufficient to sustain megakaryocytic cell growth. Erickson-Miller et al 42 reported that FBS can inhibit CFU-Meg colony formation because of the presence of inhibitory factors such as TGF- and/or platelet products. As CM also contains FBS, the difference between the two media could be explained by the presence of extracellular matrix components. To support this hypothesis, other studies have shown that glycosaminoglycans, in the presence of cytokines, are able to enhance in vitro megakaryocytopoiesis. 43,44 However, although indispensable to CD61 + cell emergence, CM is not sufficient to induce terminal differentiation of the megakaryocytic lineage. Fully developed megakaryocytes were rarely observed in these culture conditions and CD61 + cells showed a maturation delay. This suggests that full maturation of megakaryocytes in vitro requires other factors such as exogenous thrombopoietin. 45 The optimal cytokine combination for expansion remains a matter of debate. We investigated the impact on expansion of the addition of two cytokines G-CSF and MIP-1 to a basic cocktail consisting of IL-3, IL-6, IL-11 and SCF. In published expansion protocols, the addition of G-CSF to a cytokine combination gave contradictory results, being either beneficial 25

10 744 In vitro expansion of CD34 + cell subsets in conditioned medium or detrimental 26 to progenitor expansion. With our culture conditions, addition of G-CSF significantly improved CFU-GM and HPP-CFC expansion from CD34 + DR cells and was not detrimental to LTC-IC maintenance. More surprisingly, G-CSF added in CM also improved the emergence of CD61 + cells from this cell subpopulation only, supporting the notion that G-CSF acts at the level of a pluripotent progenitor cell. 24 MIP- 1 has been reported to improve progenitor expansion 28 and LTC-IC maintenance 29 from primitive populations. We did not observe any significant effect of MIP-1 regardless of condition. Polymerization of the molecule 46 or unbalance towards positive regulators 47 in the presence of a mixture of IL-3, SCF, IL-6, IL-11 ± G-CSF could explain our results. Our data are in concordance with data from Soma et al 48 who showed, using the murine long-term repopulating activity assay, that MIP-1 has no significant effect on either shortterm or long-term repopulating ability. As has been shown by others, 2,49,50 cytometric analysis showed a rapid fall in CD34 antigen expression during culture while the number of progenitors as assessed by clonogenic assay progressively increased. The percentage of CD34 + cells decreases during ex vivo culture while TNC number increases. Consequently, the absolute number of CD34 + cells is at least maintained. This underscores the discrepancies between CD34 expression and clonogenicity after expansion in liquid culture and questions the usefulness of cytometric analysis in this context. In spite of high variability of expansion potential between donors as reported by Koller et al, 51 we can conclude that (1) CM provided significant advantages over medium supplemented with FBS and over non-conditioned LTBMC medium to sustain in vitro expansion of CD34 + DR cells for at least 3 weeks. (2) Added to a basic cytokine cocktail of IL- 3, SCF, IL-6 and IL-11, G-CSF improved progenitor expansion and CD61 + cell growth without being detrimental to the LTC- IC pool while MIP-1 has no significant effect in the presence of such a complex mixture of cytokines. (3) The expansion requirements of BM cells vary according to the starting cell subpopulation analysed and to the cell pool whose expansion is required. Therefore, careful adaptation of the culture medium and cytokines to the CD34 + subpopulation studied is necessary to provide reliable and effective expansion of the desired cell population. Acknowledgements PhH is a recipient of Télévie grant No (Fonds National de la Recherche Scientifique, Belgium). This work was partly supported by the Salus Sanguinis Fundation, the Loterie Nationale, the Association Contre le Cancer and the Fonds de Développement Scientifique (Université Catholique de Louvain). We wish to thank the following companies: Sandoz (S Vancayzeele), Immunex (M Widmer) and Genetics Institute (L Lisher) for providing the cytokines. We gratefully acknowledge Dr Z Berneman, Dr D Van Bockstaele and the staff of the Hematology Laboratory, UIA, Antwerp, Belgium, for their expert assistance in cell sorting. References 1 Simmons PJ, Haylock DN. Use of hematopoietic growth factors for in vitro expansion of precursor cell populations. Curr Opin Hematol 1995; 2: Haylock DN, To LB, Dowse TL, Juttner CA, Simmons PJ. Ex vivo expansion and maturation of peripheral blood CD34 + cells into the myeloid lineage. Blood 1992; 80: Srour EF, Brandt JE, Bridell RA, Grisby S, Leemhuis T, Hoffman R. Long-term generation and expansion of human primitive hematopoietic cells in vitro. Blood 1993; 81: Moore MAS. Ex vivo expansion and gene therapy using cord blood CD34 + cells. J Hematother 1993; 2: Sato N, Sawada K, Koizumi K, Tarumi T, Leko M, Yazukouchi M, Yamaguchi TA, Takhashi S, Skigushi S, Koike T. In vitro expansion of human peripheral blood CD34 + cells. Blood 1993; 82: Brandt J, Bridell A, Srour EF, Leemhuis TB, Hoffman R. Role of c- kit ligand in the expansion of human hematopoietic progenitors. Blood 1992; 79: Henschler R, Brugger W, Luft T, Frey T, Mertelsmann R, Kanz L. Maintenance of transplantation potential in ex vivo expanded CD34 + selected human peripheral blood progenitor cells. Blood 1994; 84: Petzer AL, Hogge DE, Lansdorp PM, Reid DS, Eaves CJ. Selfrenewal of primitive human hematopoietic cells (long-termculture-initiating cells) in vitro and their expansion in defined medium. Proc Natl Acad Sci USA 1996; 93: Muench MO, Firpo MT, Moore MAS. Bone marrow transplantation with interleukin-1 plus c-kit ligand ex vivo expanded bone marrow accelerates hematopoietic reconstitution in mice without the loss of stem cell ineage and proliferative potential. Blood 1993; 81: Serano F, Varas F, Bernard A, Bueren JA. Accelerated and longterm hematopoietic engraftment in mice transplanted with ex vivo expanded bone marrow. Bone Marrow Transplant 1994; 14: Peters SO, Kittler ELW, Ramshaw HS, Quesenberry PJ. Murine marrow cells expanded in culture with IL-3, IL-6, IL-11, and SCF acquire an engraftment defect in normal hosts. Exp Hematol 1995; 23: Peters SO, Kittler ELW, Ramshaw HS, Quesenberry PJ. Ex vivo expansion of murine marrow cells with interleukin-3 (IL-3), IL-6, IL-11, and stem cell factor leads to impaired engraftment in irradiated hosts. Blood 1996; 87: Yonemura Y, Ku H, Hirayama F, Souza LM, Ogawa M. Interleukin 3 or interleukin 1 abrogates the reconstituting ability of hematopoi- etic stem cells. Proc Natl Acad Sci USA 1996; 93: Brugger W, Heimfeld S, Berenson RJ, Mertelsman R, Kanz L. Reconstitution of hematopoiesis after high dose chemotherapy by autologous progenitor cells generated ex vivo. New Engl J Med 1995; 333: Williams SF, Lee WJ, Bender JG, Zimmerman T, Swinney P, Blake M, Carreon J, Schilling M, Smith S, Williams DE, Oldham D, Van Epps D. Selection and expansion of peripheral blood CD34 + cells in autologous stem cell transplantation for breast cancer. Blood 1996; 87: Sutherland HJ, Eaves CJ, Eaves AC, Dragowska W, Lansdorp PM. Characterization and partial purification of human marrow cells capable of initiating long-term hematopoiesis in vitro. Blood 1989; 74: Verfaillie C, Blakolmer K, McGlave P. Purified primitive human hematopoietic progenitor cells with long-term in vitro repopulating capacity adhere selectively to irradiated bone marrow stroma. J Exp Med 1990; 172: Verfaillie CM, Miller WJ, Boylan K, McGlave PB. Selection of benign primitive hematopoietic progenitors in chronic myelogenous leukemia on the basis of the HLA-DR antigen expression. Blood 1992; 79: Sutherland HJ, Blair A, Zapf RW. Characterization of a hierarchy in human acute myeloid leukemia progenitor cells. Blood 1996; 87: Straetmans N, Ma DDF, Herman Ph, Zenebergh A, Chang A, Deneys V, De Bruyere M, Ferrant A. Long-term culture of autologous transplanted bone marrow for acute myeloid leukaemia: evidence for an in vitro haemopoietic defect and lack of correlation with the speed of engraftment. Bone Marrow Transplant 1995; 15: Verfaillie CM. Soluble factor(s) produced by human bone marrow stroma increase cytokine-induced proliferation and maturation of